# alibaba/graph-learn

An Industrial Graph Neural Network Framework

Repository: https://github.com/alibaba/graph-learn
Canonical: https://ross.abutalabs.com/products/graph-learn
Language: C++
License: Apache-2.0
License Family: permissive
Topics: gnn, aligraph, graphlearn, tensorflow, pytorch, graph, graph-neural-networks, gnn-framework, dynamic-graph, training, dynamic-graph-service, gnn-inference
Last push: 2025-07-04T08:52:27+00:00

## Health v2 (maintenance only)
Score: 36/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 30, release rhythm 8, longevity 100
- inputs: {"age_days": 2351, "days_push": 425, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1341, forks 266 (observed 2026-08-28T04:04:26.135134+00:00)

## What it is
Graph-Learn (formerly AliGraph) is a distributed framework for developing and applying large-scale graph neural networks, with a training library compatible with TensorFlow and PyTorch and a dynamic graph online inference service. It provides Python and C++ graph sampling interfaces with a gremlin-like GSL, and is used in production at Alibaba for search recommendation, security, and knowledge graphs.

## Use cases
- train graph neural networks on large-scale graphs
- distributed GNN model training with TensorFlow or PyTorch
- real-time graph sampling for online GNN inference
- incremental training on streaming dynamic graphs
- build recommendation models using graph embeddings
- run GNN models for fraud and network security detection
- serve GNN predictions with low P99 latency

## When to choose
- you need industrial-scale distributed GNN training
- your graph data updates in real time and you need online inference
- you want a sampling framework compatible with both TensorFlow and PyTorch
- you need low-latency sampling guarantees for serving GNNs

## When to avoid
- you only need small-scale graph analytics without neural networks
- you want a pure PyTorch-native GNN library without a distributed service layer
- your project needs Windows or macOS support
- you need a lightweight library for simple graph algorithms

## Facets
- artifact type: framework
- maturity: active
- function: machine-learning, deep-learning, llm-training, rag
- domain: machine-learning, deep-learning, large-language-models, microservices
- platform: python, cpp
- tags: gnn, graph-neural-networks, graph-sampling, tensorflow, pytorch, dynamic-graph, graph-inference, alibaba, linux, docker

## Member repositories
- alibaba/graph-learn (main) score 36

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:26.135134+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:43:43.658775+00:00, confidence not recorded.
  - readme: https://github.com/alibaba/graph-learn (fetched 2026-08-28T04:04:26.135134+00:00, sha bd2fde1926a4)
- Data as of 2026-08-30T08:39:29.467469+00:00.